This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TPS7B84-Q1: PSRR measurement

Part Number: TPS7B84-Q1

Hi team,

When the PSRR of TPS7B84-Q1 was measured, the following results were obtained.
This does not match the data sheet. What could be the cause?

  

The measurement setup and circuit conditions are as follows.

Vin=13.5V, Vout=5.0V, Vripple≈100mV, Cin=10μF, Cout=1μF

Regards,

Max Fujimoto

 
 
  • Hi Max,

    The difference is likely the input and output caps. Most of our PSRR measurements are performed with no input cap (as long as the device does not oscillate itself with no input cap) so that the oscillator can actually drive the input with a clean signal when the frequency starts getting large. The input cap will act as a filter and will attenuate the signal before it is measured at the output and the result is that the PSRR appears to be higher than it actually is. The output cap they used is also 10X smaller than what is used in the datasheet plot, which will reduce the PSRR at higher frequencies, which is why the PSRR in their measurement roles off faster.

    Regards,

    Nick

  • Hi Nick,

    Thank you for your advice. I removed Cin and measured again under the following conditions.

    Vin=13.5V, Vout=5.0V, Vripple≈100mVCout=10μF, Iout=100mA

    However, the behavior between 5kHz and 500kHz still does not match the data sheet. Are there any other possible factors? (It seems that it is the S/N ratio limit of the oscillator that the gain drops sharply toward 10MHz.)

     

    Regards,

    Max Fujimoto

  • Hi Max,

    Have you tried using a resistive load instead of a current source? I'm not sure if that would make a difference but I would be curious to see. 

    The Vripple may have an effect as well. 100mV is a somewhat large signal; can you try 10mV to see if the PSRR curve changes? 

    Regards,

    Nick

  • Hi Nick,

    A resistive load means like an electronic load in constant resistance mode?

    I tried with smaller Vripple, but there was no significant change in the waveform. There is a noticeable decrease in gain around 100kHz.

    Could you tell me if there is any difference in the setup for measuring PSRR? I would like to know the signal superimposition method that you usually measure PSRR.

    Regards,

    Max 

  • Hi Max,

    No, I mean a resistor for the load. We try avoiding using e-loads for tests like this because the control loops can interact and cause unexpected behavior. 

    Are you using the official EVM for this measurement? If not, can you share the schematic and layout of the board you're using?

    Thanks,

    Nick

  • Hi Nick,

    OK. I think there was a variable resistor in the lab, so I will experiment with it.

    The EVM is the official one.

    Thanks,

    Max

  • Hi Max,

    I see. In that case, try reducing the setup parasitics as much as possible. What I'm thinking is that if you are using long wires to connect everything, the parasitic inductance, if large enough, can start reducing PSRR at higher frequencies.

    Regards,

    Nick

  • Hi Nick,

    Here is the result that I measured with a resistor, not e-load. 

    Vin=13.5V, Vout=5.0V, Vripple≈100mVCout=10μF, Iout=100mA (with a resistor)

    Compared to the previous result with e-load, there isn't significant change. 

    I also tried 10mV ripple, but the waveform does not seem to change except for some noise. 

    Vin=13.5V, Vout=5.0V, Vripple≈10mVCout=10μF, Iout=100mA (with a resistor)

    What I care most about is the drop at 100kHz. I think this part is unlikely to change even if the parasitic inductance is differed. What do you think this is due to?

    Thanks,

    Max

  • Hi Max,

    I think you're probably correct in the ~100kHz range. The only other thing I can think of is coupling from the input to the output. If there is an input cap (I know you removed it), noise can couple through the input cap to ground and back through the output cap, which lowers the measured PSRR. With that said, I'm not seeing any obvious coupling paths on the EVM. 

    I will be out of office tomorrow until Monday 6/20. If you can wait until then, I can go in and try myself to reproduce the PSRR plot seen in the datasheet. Can you wait until then?

    Thanks,
    Nick

  • Hi Nick,

    No problem. I really appriciate it. I'm going to wait for your results. 

    Thanks,

    Max

  • Hi Max,

    Sounds good. Thanks for being patient.

    Regards,

    Nick

  • Hi Max,

    I developed COVID-19 symptoms yesterday and so I will not be able to get into the office to run this test for you until at least sometime next week. Very sorry for the inconvenience. 

    Regards,

    Nick

  • Hi Nick,

    I’m sorry to hear that. Feel better soon!

    Regards.

    Max

  • Hi Max,

    Just wanted to give you an update. I tried reproducing the PSRR data and I was having some issues. I think it was setup-related. I will be back in the office next week starting Tuesday. I'll try again then.

    Thanks,

    Nick

  • Hi Nick,

    That would be fine. Thank you for your update.

    Regards,

    Max

  • Hi Max,

    Another update. I did not have time until today to work more on this and I need to order more parts. I've submitted an order and I expect it to take 1-2 weeks to get to me. Very sorry for the inconvenience and that it's taking so long to get this resolved.

    Regards,

    Nick

  • Hi Nick,

    If you have any updates, please reply at your convenience.

    Regards,

    Max

  • Hi Max,

    Very sorry for the delay. I have the parts on my desk now but I haven't had the chance to go back in and test them. I'll try getting back in the lab and provide some results by the end of this week.

    Thanks,

    Nick

  • Hi Max,

    Very sorry but I was not able to get to this. I am out of office tomorrow so I will have to do this next week. 

    Regards,

    Nick